WEBVTT

00:00:00.000 --> 00:00:01.890 align:middle line:90%
Hello, I'm Rahul Agarwal.

00:00:01.890 --> 00:00:03.360 align:middle line:90%
I'm senior engineer at RITMO.

00:00:03.360 --> 00:00:08.010 align:middle line:84%
I work with eye tracking, fMRI,
a motion capture system, EEG,

00:00:08.010 --> 00:00:10.050 align:middle line:90%
and intracranial EEG.

00:00:10.050 --> 00:00:11.640 align:middle line:90%
And I'm Laura Bishop.

00:00:11.640 --> 00:00:13.290 align:middle line:90%
I'm a postdoc at RITMO.

00:00:13.290 --> 00:00:15.990 align:middle line:84%
And I work with eye
tracking, and pupillometry,

00:00:15.990 --> 00:00:19.410 align:middle line:84%
and motion capture, usually
with music ensembles.

00:00:19.410 --> 00:00:20.430 align:middle line:90%
Welcome to week three.

00:00:20.430 --> 00:00:24.150 align:middle line:84%
Here, we'll be discussing
about eye tracking, advantages,

00:00:24.150 --> 00:00:28.080 align:middle line:84%
disadvantages of eye trackers,
both mobile and stationary eye

00:00:28.080 --> 00:00:29.250 align:middle line:90%
trackers.

00:00:29.250 --> 00:00:31.350 align:middle line:84%
We'll be giving
some applications

00:00:31.350 --> 00:00:34.380 align:middle line:84%
regarding eye trackers, as well
as understanding how to choose

00:00:34.380 --> 00:00:38.610 align:middle line:84%
and how to make choice for
your application of the study

00:00:38.610 --> 00:00:40.140 align:middle line:90%
for the eye trackers.

00:00:40.140 --> 00:00:44.040 align:middle line:84%
So can you tell us a little bit
about some of the advantages

00:00:44.040 --> 00:00:46.920 align:middle line:84%
and disadvantages of using
mobile eye tracking instead

00:00:46.920 --> 00:00:47.730 align:middle line:90%
of stationary?

00:00:47.730 --> 00:00:50.430 align:middle line:84%
The most important advantage
of mobile eye tracker

00:00:50.430 --> 00:00:52.110 align:middle line:90%
is portability.

00:00:52.110 --> 00:00:55.230 align:middle line:84%
Mobile eye trackers offer
advantage of portability

00:00:55.230 --> 00:00:56.940 align:middle line:90%
over screen-based eye trackers.

00:00:56.940 --> 00:01:02.280 align:middle line:84%
These eye trackers are used to
get the gaze behaviour in real

00:01:02.280 --> 00:01:03.120 align:middle line:90%
world environment.

00:01:03.120 --> 00:01:06.780 align:middle line:84%
This eye trackers can be worn by
anyone and usually helpful when

00:01:06.780 --> 00:01:09.580 align:middle line:84%
you are collecting data
in realistic world,

00:01:09.580 --> 00:01:13.800 align:middle line:84%
such as in sports facilities
or shopping malls,

00:01:13.800 --> 00:01:15.900 align:middle line:90%
or superstores, et cetera.

00:01:15.900 --> 00:01:18.180 align:middle line:84%
The ideal condition for
using these type of eye

00:01:18.180 --> 00:01:21.690 align:middle line:84%
trackers, either if you have
head movements, a lot of head

00:01:21.690 --> 00:01:24.270 align:middle line:84%
movements or if the
subject is moving around

00:01:24.270 --> 00:01:25.890 align:middle line:90%
in the real world.

00:01:25.890 --> 00:01:29.850 align:middle line:84%
And the cons for this eye
tracker is it's very expensive.

00:01:29.850 --> 00:01:33.240 align:middle line:84%
The technological setup
and data analysis,

00:01:33.240 --> 00:01:35.880 align:middle line:90%
it's quite labour intensive.

00:01:35.880 --> 00:01:39.600 align:middle line:84%
This eye tracker must be used
in very controlled environment,

00:01:39.600 --> 00:01:42.030 align:middle line:84%
where lighting condition
is not affecting

00:01:42.030 --> 00:01:44.040 align:middle line:90%
the performance of eye tracker.

00:01:44.040 --> 00:01:45.900 align:middle line:84%
So you have to do
this everything

00:01:45.900 --> 00:01:51.000 align:middle line:84%
in a very controlled environment
for mobile eye trackers.

00:01:51.000 --> 00:01:52.770 align:middle line:84%
Can you tell me
some pros and cons

00:01:52.770 --> 00:01:54.750 align:middle line:90%
for stationary eye trackers?

00:01:54.750 --> 00:01:56.040 align:middle line:90%
Yeah, well the--

00:01:56.040 --> 00:01:59.040 align:middle line:84%
I guess the main pros of
using stationary eye tracking

00:01:59.040 --> 00:02:01.770 align:middle line:90%
is that it is very controlled.

00:02:01.770 --> 00:02:05.880 align:middle line:84%
You're able to set up the eye
tracker in a lab like this,

00:02:05.880 --> 00:02:08.070 align:middle line:84%
where the conditions
are very stable

00:02:08.070 --> 00:02:10.930 align:middle line:84%
and everything-- the lighting
is set to a certain level

00:02:10.930 --> 00:02:12.550 align:middle line:90%
and so forth.

00:02:12.550 --> 00:02:17.252 align:middle line:84%
And therefore, you're able
to get very precise data.

00:02:17.252 --> 00:02:18.960 align:middle line:84%
So if you're looking
at eye tracking data

00:02:18.960 --> 00:02:22.470 align:middle line:84%
and you want to look at
the details of exactly

00:02:22.470 --> 00:02:25.030 align:middle line:84%
where somebody is looking
very precisely in the saccades

00:02:25.030 --> 00:02:30.260 align:middle line:84%
and so forth, then it's better
to use stationary eye tracking.

00:02:30.260 --> 00:02:34.460 align:middle line:84%
But on the other hand,
there's some disadvantages.

00:02:34.460 --> 00:02:37.670 align:middle line:90%
It's not portable, as you said.

00:02:37.670 --> 00:02:41.690 align:middle line:84%
It's very restrictive
in terms of the extent

00:02:41.690 --> 00:02:44.120 align:middle line:84%
to which the participants
can move around.

00:02:44.120 --> 00:02:47.360 align:middle line:84%
And it also constrains
where they're

00:02:47.360 --> 00:02:51.650 align:middle line:84%
able to look, usually to
a small computer screen.

00:02:51.650 --> 00:02:55.250 align:middle line:84%
So it only captures a
portion of the visual field.

00:02:55.250 --> 00:02:58.760 align:middle line:84%
And it's also quite
boring for participants

00:02:58.760 --> 00:03:01.320 align:middle line:84%
to do these kinds
of experiments.

00:03:01.320 --> 00:03:05.210 align:middle line:84%
And for spending a long period
of time looking at a screen

00:03:05.210 --> 00:03:08.330 align:middle line:84%
with not much going on, they
might get a little tired

00:03:08.330 --> 00:03:10.790 align:middle line:90%
and a little inattentive.

00:03:10.790 --> 00:03:13.010 align:middle line:84%
So the stationary eye
tracker would then

00:03:13.010 --> 00:03:15.020 align:middle line:84%
be the best option
for situations

00:03:15.020 --> 00:03:18.470 align:middle line:84%
where you're looking,
really, for a lot of detail

00:03:18.470 --> 00:03:21.950 align:middle line:84%
either in the gaze or
in the pupil tracking.

00:03:21.950 --> 00:03:24.110 align:middle line:84%
Yes, probably that
would be the best one

00:03:24.110 --> 00:03:27.680 align:middle line:84%
to use for if you're
looking at the applications

00:03:27.680 --> 00:03:29.060 align:middle line:90%
you're looking for, yeah.

00:03:29.060 --> 00:03:32.780 align:middle line:84%
And the mobile eye tracker
would be best in situations

00:03:32.780 --> 00:03:35.900 align:middle line:84%
where a little bit
less control is OK.

00:03:35.900 --> 00:03:37.890 align:middle line:90%
Yep, if your head is moving--

00:03:37.890 --> 00:03:40.460 align:middle line:84%
if your head is moving and
if your desire your subject

00:03:40.460 --> 00:03:42.740 align:middle line:84%
will be moving around
in an environment.

00:03:42.740 --> 00:03:45.170 align:middle line:84%
So you can use the mobile
eye tracking system.

00:03:45.170 --> 00:03:48.650 align:middle line:84%
And this kind of system then
also allows for a lot more

00:03:48.650 --> 00:03:50.250 align:middle line:90%
realistic behaviour.

00:03:50.250 --> 00:03:56.780 align:middle line:84%
So if you have a type of
activity, like sports or music,

00:03:56.780 --> 00:03:59.690 align:middle line:84%
where a person would be very
constrained and actually

00:03:59.690 --> 00:04:01.790 align:middle line:84%
their attention would
also change from normal

00:04:01.790 --> 00:04:04.850 align:middle line:84%
if they were sitting in a
seat with their head fixed,

00:04:04.850 --> 00:04:07.110 align:middle line:84%
then this would be
a better option.

00:04:07.110 --> 00:04:10.970 align:middle line:84%
Yeah, that will be
the better option

00:04:10.970 --> 00:04:13.040 align:middle line:84%
if you have somebody
an environment,

00:04:13.040 --> 00:04:16.690 align:middle line:84%
such as you're testing in
sports facilities, or malls,

00:04:16.690 --> 00:04:19.829 align:middle line:84%
superstores, or doing
that kind of stuff.

00:04:19.829 --> 00:04:23.890 align:middle line:84%
So some of the applications
that we use eye tracking for,

00:04:23.890 --> 00:04:26.730 align:middle line:84%
you've just given a
couple of examples.

00:04:26.730 --> 00:04:28.760 align:middle line:84%
So some examples can
be-- you can use it

00:04:28.760 --> 00:04:35.450 align:middle line:84%
in virtual gaming, as well
as virtual reality, yeah.

00:04:35.450 --> 00:04:40.070 align:middle line:84%
And the extended-- you can
see the extended applications

00:04:40.070 --> 00:04:42.830 align:middle line:84%
can be heavy
industrial companies,

00:04:42.830 --> 00:04:45.290 align:middle line:84%
where people have
to figure out what

00:04:45.290 --> 00:04:47.220 align:middle line:90%
is the hazard in the workplace.

00:04:47.220 --> 00:04:49.400 align:middle line:84%
So that can be
used to figure out

00:04:49.400 --> 00:04:52.340 align:middle line:84%
the research where is
the actual hazard placed

00:04:52.340 --> 00:04:55.400 align:middle line:90%
in the industrial area.

00:04:55.400 --> 00:04:59.000 align:middle line:84%
And some of the applications
that I use, I guess,

00:04:59.000 --> 00:05:03.140 align:middle line:84%
in my area, since I'm interested
in music, I do a lot of work

00:05:03.140 --> 00:05:06.560 align:middle line:84%
looking in live
performance situations

00:05:06.560 --> 00:05:08.330 align:middle line:84%
and seeing where
people are looking

00:05:08.330 --> 00:05:10.790 align:middle line:90%
when they're playing music.

00:05:10.790 --> 00:05:13.100 align:middle line:84%
Other musical
applications would be

00:05:13.100 --> 00:05:17.090 align:middle line:84%
seeing how people's
gaze changes as they're

00:05:17.090 --> 00:05:19.370 align:middle line:84%
reading a musical score
or as they're watching

00:05:19.370 --> 00:05:21.320 align:middle line:90%
a performance where they look.

00:05:21.320 --> 00:05:24.020 align:middle line:84%
Yeah, so how do
you actually choose

00:05:24.020 --> 00:05:27.050 align:middle line:84%
whether to use mobile eye
tracker or stationary eye

00:05:27.050 --> 00:05:27.590 align:middle line:90%
tracker?

00:05:27.590 --> 00:05:29.540 align:middle line:90%
How do you differentiate?

00:05:29.540 --> 00:05:32.900 align:middle line:84%
Well, I guess it would depend on
what your research question is.

00:05:32.900 --> 00:05:36.830 align:middle line:84%
And if it's more important
for your research question

00:05:36.830 --> 00:05:42.050 align:middle line:84%
that the people are
behaving very naturally,

00:05:42.050 --> 00:05:44.450 align:middle line:84%
and they're doing an activity
that involves movement

00:05:44.450 --> 00:05:47.270 align:middle line:84%
and behaving naturally is
going to affect the way

00:05:47.270 --> 00:05:50.090 align:middle line:84%
that they attend and
they carry out the task,

00:05:50.090 --> 00:05:53.270 align:middle line:84%
then it's better to choose
mobile eye tracking.

00:05:53.270 --> 00:05:55.280 align:middle line:84%
But if your research
question demands

00:05:55.280 --> 00:05:57.800 align:middle line:84%
a lot of precision
in your measures,

00:05:57.800 --> 00:06:00.350 align:middle line:84%
then it's better to
choose stationary.

00:06:00.350 --> 00:06:02.810 align:middle line:84%
Yeah, what are the
best lab conditions

00:06:02.810 --> 00:06:06.428 align:middle line:84%
for testing the eye
tracking experiments?

00:06:06.428 --> 00:06:07.970 align:middle line:84%
Well, for the
stationary eye tracker,

00:06:07.970 --> 00:06:10.310 align:middle line:84%
we normally use
this kind of room.

00:06:10.310 --> 00:06:14.930 align:middle line:84%
So this is our stationary
eye tracker here.

00:06:14.930 --> 00:06:17.300 align:middle line:90%
It's a very calm and quiet room.

00:06:17.300 --> 00:06:20.850 align:middle line:84%
We keep the lighting
very stable.

00:06:20.850 --> 00:06:23.030 align:middle line:84%
We have lights
coming from overhead.

00:06:23.030 --> 00:06:25.490 align:middle line:84%
So they're not shining
into the light--

00:06:25.490 --> 00:06:28.010 align:middle line:84%
shining into the eyes
of the participants.

00:06:28.010 --> 00:06:33.210 align:middle line:84%
We have light coloured walls
with nothing much on them.

00:06:33.210 --> 00:06:35.900 align:middle line:84%
And it's also important
if there's any windows,

00:06:35.900 --> 00:06:38.790 align:middle line:90%
like we have here--

00:06:38.790 --> 00:06:43.050 align:middle line:84%
Yeah, so that's how you
decide that's the best

00:06:43.050 --> 00:06:46.310 align:middle line:90%
condition for the eye tracking.

00:06:46.310 --> 00:06:49.730 align:middle line:84%
Yeah, in the room where you're
doing stationary eye tracking,

00:06:49.730 --> 00:06:52.100 align:middle line:84%
then you are able
to set the room up

00:06:52.100 --> 00:06:54.260 align:middle line:84%
like this in very
controlled conditions.

00:06:54.260 --> 00:06:56.690 align:middle line:84%
And then if you're going
to be taking the mobile eye

00:06:56.690 --> 00:06:59.840 align:middle line:84%
trackers out, you might
not have that much control

00:06:59.840 --> 00:07:01.350 align:middle line:90%
at all over the conditions.

00:07:01.350 --> 00:07:04.790 align:middle line:84%
But if possible,
you would still try

00:07:04.790 --> 00:07:08.510 align:middle line:84%
to set up your space so that
there's no windows shining,

00:07:08.510 --> 00:07:10.610 align:middle line:90%
bright lights.

00:07:10.610 --> 00:07:12.327 align:middle line:84%
I don't know, if I'm
looking at musicians

00:07:12.327 --> 00:07:13.910 align:middle line:84%
and they're on the
stage, for example,

00:07:13.910 --> 00:07:17.360 align:middle line:84%
I would rather use kind of
constant lighting, no stage

00:07:17.360 --> 00:07:18.510 align:middle line:90%
lighting.

00:07:18.510 --> 00:07:22.430 align:middle line:84%
Directly focusing on the
face, that's what you mean.

00:07:22.430 --> 00:07:26.450 align:middle line:84%
And how does the eye tracker
works with different people,

00:07:26.450 --> 00:07:32.920 align:middle line:84%
for say, if you different face
shape, or different eye colour,

00:07:32.920 --> 00:07:35.770 align:middle line:84%
or people who have
visual impairment?

00:07:35.770 --> 00:07:37.690 align:middle line:90%
How does that work with them?

00:07:37.690 --> 00:07:40.720 align:middle line:84%
Well, the eye trackers are
meant to work on most people.

00:07:40.720 --> 00:07:43.267 align:middle line:84%
And they do work well
with most people.

00:07:43.267 --> 00:07:45.100 align:middle line:84%
Every once in a while
you have some problems

00:07:45.100 --> 00:07:47.860 align:middle line:90%
with the mobile eye tracker.

00:07:47.860 --> 00:07:53.980 align:middle line:84%
Because in this case, the
structure of the whole system

00:07:53.980 --> 00:07:54.970 align:middle line:90%
is--

00:07:54.970 --> 00:07:57.230 align:middle line:84%
it needs to fit your
face in a certain way.

00:07:57.230 --> 00:07:59.840 align:middle line:84%
So depending on the
shape of your nose,

00:07:59.840 --> 00:08:02.560 align:middle line:84%
for example, the
position of your eyes,

00:08:02.560 --> 00:08:04.240 align:middle line:84%
every once in a while,
you find someone

00:08:04.240 --> 00:08:09.200 align:middle line:84%
that the eye tracker just does
not work very well for them.

00:08:09.200 --> 00:08:12.190 align:middle line:84%
And depending on the
system you're using,

00:08:12.190 --> 00:08:17.290 align:middle line:84%
it can be important
that the participants

00:08:17.290 --> 00:08:20.270 align:middle line:90%
have eyes that converge.

00:08:20.270 --> 00:08:23.770 align:middle line:84%
And so if there's somebody
who has a condition where

00:08:23.770 --> 00:08:27.280 align:middle line:84%
one of their eyes is not
converging with the other,

00:08:27.280 --> 00:08:29.020 align:middle line:84%
then that might not
be-- it might not

00:08:29.020 --> 00:08:30.040 align:middle line:90%
be possible to record.

00:08:30.040 --> 00:08:32.020 align:middle line:84%
But there's also
some systems that

00:08:32.020 --> 00:08:34.120 align:middle line:90%
record the two eyes separately.

00:08:34.120 --> 00:08:35.990 align:middle line:84%
And in that case,
it's not a problem.

00:08:35.990 --> 00:08:38.535 align:middle line:90%
OK.

00:08:38.535 --> 00:08:40.520 align:middle line:90%
That's the end.

00:08:40.520 --> 00:08:52.000 align:middle line:90%